Abstract:

Abstract Purpose. The classic algorithms for finding the shortest path on the graph that underlie existing routing protocols, which are now used in computer networks, in conditions of constant change in network traffic can not lead to the optimal solution in real time. Methodology. To determine the optimal route in the computer network, the program model «MLP 34-2-410-34» was developed in Python using the TensorFlow framework, which allows the following steps to be performed: sample generation (random or balanced ); the creation of a neural network, the input of which is an array of bandwidth channels of the computer network, as a resultant array of signs of the use of the appropriate communication channel in the formation of the route in the computer network; training and testing of the neural network in the appropriate samples. Findings. Neural network configuration 34-2-410-34 with activation functions of ReLU and Leaky-ReLU in a hidden layer and a linear activation function in the output layer learns from Adam algorithm, which is a combination of Adagrad, RMSprop algorithms and stochastic gradient descent with inertia, the fastest on of all volumes of the training sample, the rest of the others are overwhelmed by the conversion and reaches the value of the error at 0.0024 on the control voter and 86 % returns the optimal path. Originality. The study of the parameters of the neural network on the basis of the calculation of the average harmonic with different activation functions (Linear, Sigmoid, Tanh, Softplus, ReLU, L-ReLU) on training samples of different volumes (140, 1400, 14000, 49000 examples) and various training algorithms Neural Network (BGD, MB SGD, Adam, Adamax, Nadam). Practical value. The use of a neural model, the input of which gives the value of bandwidth channels, will allow in real time to determine the optimal route in the computer network.

Keywords: